In this paper, we present a framework for an adaptive and risk-aware robot motion planning and control, and discuss how such a framework could handle uncertainty in human workers' actions and robot localization. We build on our prior investigation, where we describe how uncertainty in human actions can be modeled using the entropy rate in a Markov Decision Process. We then describe how we can incorporate this model of uncertainty into simulations of a simple collaborative system, involving one human worker and one robotic assistant, to produce risk-aware robot motions. Next, we highlight the difficulties associated with localization uncertainty in a space environment and describe how we can incorporate this uncertainty into an adaptive system as well. Expected advantages of an adaptive system are described, including increases in overall efficiency due to reductions in idle time, increases in concurrent motion, faster task execution, as well as subjective improvements in the worker's satisfaction with the assistant and reduced worker stress and fatigue. A pilot experiment designed to evaluate the benefits of introducing risk-aware motion planning is described. It is found that human-robot teams in which the robot utilizes risk-aware motion planning show on average 24% more concurrent motion and execute the task 13% faster, while simultaneously improving safety by having a 19.9% larger mean separation distance between the human and robot workers. Finally, possible future system developments and user studies are discussed.